Electrochemical Reduction of Carbon Dioxide over CNT-Supported Nanoscale Copper Electrocatalysts

This paper presents the experimental investigation of copper loaded carbon nanotubes (CNTs) electrocatalysts for the electrochemical reduction of carbon dioxide. The electrocatalysts were synthesized by homogeneous deposition precipitation method (HDP) using urea as precipitating agent. The prepared catalysts were characterized by TEM, SEM, XRD, XPS, BET, and FTIR for their morphology and structure. Characterization results confirm the deposition of Cu nanoparticles (3–60 nm) on CNTs. Linear sweep voltammetry (LSV) and chronoamperometry (CA) were used to investigate the activity of the as-prepared catalysts for the electrochemical reduction of carbon dioxide. The electrocatalysts reduced CO 2 with high current density in the potential range 0∼−3V versus SCE (standard calomel electrode). Among all catalysts tested, 20wt. % copper loaded CNTs showed maximum activity. Gas chromatograph with TCD was used to analyze liquid phase composition. The faradaic efficiency for methanol formation was estimated to be 38.5%.


Introduction
Scientific evidences are quite persuasive that the global warming has direct link with the increased level of greenhouse gases in atmosphere.Carbon dioxide (CO 2 ) is one of the major greenhouse gases.Its concentration stands at 390 ppm in 2011 in comparison to 350 ppm in 1960 [1,2].Substantial steps need to be taken to meet the Kyoto protocol which limits the average earth temperature rise to 2 ∘ C [2,3].Fossil fuel combustion for energy generation and transportation purposes is regarded as one of the main sources of CO 2 .Various strategies, including shifting fuel utilization pattern from fossil fuel to renewable energy, increasing the efficiency of engines, carbon dioxide sequestration, and chemical utilization, have been proposed to combat the everincreasing concentration of CO 2 .Of all the strategies, direct use of CO 2 as feed for producing high value chemicals is particularly attractive.Electrochemical reduction of carbon dioxide (ERC) is one of the most promising methods for CO 2 mitigation because (A) it uses CO 2 , a greenhouse gas, and water as feed, (B) it produces high economic value products like methanol, methane, ethylene, formic acid, and synthesis gas, (C) it operates at room temperature and pressure, and (D) it provides possibility of using renewable sources of energy for the process.ERC have been carried out mostly in aqueous and nonaqueous medium.ERC suffers from the requirement of high electrode overpotential and low resultant current density [4,5].At high potential, if the reaction is carried out in aqueous phase, hydrogen evolution reaction competes with the ERC [6].This seriously decreases the energy efficiency of the process.Product selectivity in electrochemical reduction of CO 2 depends on many factors, such as concentration of CO 2 , electrode potential, temperature, electrocatalyst material, and electrolyte solution (aqueous or nonaqueous) [5].Use of electrocatalysts has been suggested to lower the potential requirement and increase the current density [7].Moreover, product distribution can be tuned by choosing the appropriate electrocatalyst.Since Hori's work in 1983, most of the metals in periodic table and their oxides have been tried as an electrocatalyst to achieve reasonable activity for ERC at low potential with high current density [8].Copper has been found to be, till date, the most promising electrocatalyst for ECR [9].It shows medium hydrogen overvoltage with reaction products being methanol, ethane, methane, and ethylene [5,10].Copper metal [11,12], fine copper particles [13], copper oxide [14][15][16][17], and modified copper oxides have been studied extensively [18].
Electrode materials have been, mostly, employed in the form of foil, plate, and rotating disc, bed of particles, or gas diffusion electrodes [19][20][21].However, very limited studies have been carried out on metal supported on porous conducting materials as electrocatalysts for ERC.It has been suggested that ERC involves adsorption of the reactant species on the active material [4]; dispersing the metal on a high surface area support could well result in increased activity of the electrocatalyst.In addition, the use of an inert support enhances C-C chain growth probability, and hydrocarbon formation is favored by the presence of micropores since mass-transfer resistance is low and residence time is improved [22].
Yamamoto et al. [23] examined various metals, including copper, electrocatalyst supported on activated carbon fibers.These catalysts showed high current density and selectivity compared to the planar electrodes.
Carbon nanotubes (CNTs) have unique properties such as uniform and straight pores, nearly uniform meso-and macropore structure, inert surface properties, resistance to acid and base environments, and their electrical conductivity [24].Their special and steady structural characteristics and morphology are quite suitable for use as support materials especially for electrochemical applications.CNTs have emerged as a popular support material for electrocatalysts for fuel cells [24].But, to the best of our knowledge, only one study has been reported so far with CNTs as catalyst support material for ERC.Gangeri et al. used platinum and iron supported CNTs as electrocatalyst to produce long chain oxygenates due to "space restriction" effect [22].The present work was undertaken with the objective of studying the catalytic activity of nanoscale copper electrocatalysts supported on CNTs for ERC.Copper loaded CNTs catalysts were prepared by homogenous deposition precipitation (HDP) method.The morphology, structural properties, and composition of the prepared Cu/CNT nanocatalysts were characterized by using X-ray diffraction (XRD), N 2 adsorption/desorption isotherm, energy dispersive X-ray spectroscopy (EDX), transmission electron microscopy (TEM), scanning electron microscopy (SEM), and X-ray photoelectron spectra (XPS).The effect of Cu loading on the catalytic activity of Cu/CNTs for ERC was investigated in an aqueous phase divided cell.[25,26].

Experimental
Copper loaded CNTs catalysts were prepared by homogeneous deposition precipitation method (HDP) using urea as precipitating agent [27].In a typical synthesis procedure, required amount of copper nitrate pentahydrate (Merck) was added in a constantly stirred suspension made by suspending 0.8 g functionalized CNTs in 200 mL deionized water.After heating the mixture to 90 ∘ C, 30 mL of 0.42 M aqueous urea solution was added dropwise.The suspension was maintained at 90 ∘ C with vigorous stirring for deposition to take place.After deposition for 6 hrs, the slurry was cooled to room temperature, centrifuged, thoroughly washed with deionized water, and dried at 110 ∘ C for 5 hrs.The powder was calcined in a tubular furnace at 400 ∘ C for 3 hrs under argon (Ar) flow of 100 mL/min.The catalyst was finally reduced to convert the catalyst into metallic form by a gas mixture (10% H 2 in Ar) flowing at 100 mL/min up to 400 ∘ C for 3 hrs at a heating rate of 5 ∘ C/min.By using the similar procedure with different amount of copper nitrate, different loadings of copper loaded on CNTs were prepared.The prepared catalysts were denoted as %-Cu/CNT, where  represents the percent weight content of Cu loading by weight (10,20,30,40,50, and 60%).

Characterization of Catalyst. Specific surface areas (BET)
and pore volume of the synthesized materials/catalysts were determined by N 2 adsorption using a Micromeritics model ASAP 2000 analyzer.Prior to the measurements, the samples were degassed at 200 ∘ C under nitrogen flow for 5 hrs in order to remove moisture completely.Physical adsorption of N 2 was carried out in a liquid nitrogen bath maintained at 77 K temperature.
X-ray diffraction patterns of catalyst were collected in order to characterize the crystalline phases of the catalysts and crystal size.XRD analysis of the prepared catalyst was performed by a Rigaku diffractometer (D/MAX-IIIA, 3 kW) using CuK  radiation (40 kV, 30 mA,  = 0.1543 nm) to record the diffraction patterns between 20 ∘ ≥ 2 ≥ 80 ∘ with a scan rate of 6 ∘ min −1 .
Transmission electron microscopy (TEM) images were taken to determine the metal dispersion on support material along with particle size of loaded metal.An ultrahigh resolution FETEM (JEOL, JEM-2100F) at an accelerating voltage of 200 kV was employed to capture the images of the solid samples.The morphologies of the support and catalysts were studied by using a scanning electron microscope (JEOL JSM-6460LV) operated at 20 kV equipped with energy dispersive X-ray (EDX).EDX was carried out to find the composition of catalyst samples.Sample for both TEM and SEM was prepared by one milligram of the catalyst sonicated in 10 mL of ethanol for 30 min and then a drop of the suspension was placed on an aluminum foil which was mounted on the sample holder by means of a double sided adhesive copper tape.Fourier transform infrared (FTIR) spectra were obtained in transmittance mode on a Perkin-Elmer FC-16 FTIR spectrometer.Spectra were taken by making pellet of different powder samples with potassium bromide (KBr).Around 4-5 mg of sample was mixed with approximately 200 mg of KBr to prepare the pellets.The spectrum for each pellet was taken with the abovementioned spectrometer using 16 scans per sample at a resolution of 4 cm −1 over the range 4000-400 cm −1 .
X-ray photoelectron spectra (XPS) were analyzed with a V.G.Scientific ESCALAB MKII spectrometer equipped with a dual aluminum-magnesium anode X-ray gun and a 150 mm concentric hemispherical analyzer.Core level photoelectron spectra were collected on a VG scientific ESCALAB MKII spectrometer equipped with a dual aluminum-magnesium anode X-ray gun and a 150 mm concentric hemispherical analyzer using Al-Ka (h = 1486.6eV) radiation from an anode operated at 130 W.

Electrocatalytic Activity Test.
Electrochemical studies were carried out in a conventional indigenously made Htype cell with anode and cathode compartments separated by proton conducting Nafion 117 membrane.Volume of each compartment was 250 cm 3 .A 0.5 molar aqueous solution of sodium bicarbonate was used both as catholyte and anolyte.Toray carbon paper coated with catalyst was used as working electrode along with platinum plate as counter electrode.Active surface area of working electrode and counter electrode was 1.0 cm 2 and 8.0 cm 2 , respectively.Catalyst coated carbon paper was made by brush painting the catalyst ink prepared by ultrasonicating 50 mg of the catalysts along with iso-propanol and 0.82 mL of 5% Nafion solution for 30 min.Standard calomel electrode (SCE) was used as the reference electrode for the experiments.All the potentials in this study are reported with respect to SCE.Prior to the experiment, the catholyte solution was purged with high purity nitrogen for 30 min at 50 mL/min to expel dissolved air and, subsequently, high purity (99.998%) carbon dioxide was purged at 100 mL/min for 1 hr to saturate the electrolyte.The pH of the catholyte before and after CO 2 purging was 8.67 and 7.38, respectively.The pH of the catholyte did not change after purging CO 2 for one hour which indicates that one hour purging was sufficient to fully saturate the catholyte with CO 2 .Linear sweep voltammetry was carried out from 0 to −3 V with scan rate of 20 mV/s and chronoamperometry was done at −1.7 V for 3000 seconds.All electrochemical experiments were performed with PGSTAT 302 (Metrohm-Autolab, The Netherlands).Qualitative analysis of product in the catholyte was carried out by an Agilent 7890 A gas chromatograph equipped with thermal conductivity detector (TCD).Products in the gas phase were not analyzed.

X-Ray Diffraction (XRD).
The XRD analysis was carried out to investigate the diffraction structure of the Cubased catalysts supported on carbon nanotubes (CNTs) for different metal loading conditions.Figure 1 shows the X-ray diffraction pattern for the six catalyst samples with different loadings.As shown, the peaks located at 2 = 26.18∘ and 43.18 ∘ were due to the diffractions of the (002) and (100) planes of graphite-like tube wall of the CNTs [28,29].For all the supported catalysts, apart from the characteristic CNTs peaks characteristic diffraction peaks of zero valent Cu phases of (111), (200), and (220) at 2 = 43.23 ∘ , 50.10 ∘ , and 73.97 ∘ , respectively, were identified (JCPDS, number 85-1326) [30].This shows that copper was successfully deposited on CNTs by the homogenous deposition precipitation method.By comparing the peak intensities of the patterns for different Cu loading, the higher loading catalysts showed stronger peak intensity than the low loading catalysts.This revealed that the bigger the Cu particles supported on CNT, the stronger the XRD pattern.Mean Cu particle size can be determined from the characteristic Cu (111) peak by the Scherrer equation.The mean particle sizes of Cu were estimated by analyzing the line broadening of the strongest (111) diffraction peak.The mean particle sizes were calculated by the Scherrer equation,  = 0.9/ cos  where  is the X-ray wavelength (1.54 Å),  is the Bragg angle of the peak of interest, and  is the line broadening measured from the increased peak width at half height (FWHM) [29].Table 1 summarizes the analysis of the XRD results.As can be seen in Table 1, with increasing Cu loading from 10% to 60%, the crystallite size of the Cu particles increased significantly from 3.5 to around 60 nm, indicating a larger sized aggregate in the case of the higher Cu loading.

Fourier Transform Infrared (FTIR)
. FTIR spectra of functionalized and Cu coated CNTs are shown in Figure 2. The peaks which are identified at 1360, 1710, and 3402 cm −1 characterize C-O, C=O, and O=H bonds in the functionalized CNTs.Absorption peaks at 3441 cm −1 and at 1385 cm −1 correspond to the hydroxyl bands (-OH).Peaks at 1710 and  3450 cm −1 can be attributed to acidic groups like carboxyl, phenol, and lactol.Peak at 1576 cm −1 assigns C=C bond in CNTs.These observations show that the acid groups were successfully introduced in the CNTs [31].
3.1.3.X-Ray Spectroscopy.X-ray photoelectron was performed to acquire the detailed information about the chemical state of the Cu species.Figures 3(a), 3(b), and 3(c) show the XPS spectrum of 20 wt.% Cu/CNT catalyst.Spectrum in Figure 3(a) shows a strong carbon peak of C1 s at 284 eV which shows the pyrolytic carbon substrate CNT.In Figures 3(b) and 3(c) there are strong Cu peaks of Cu 2 p 3 , Cu 2 p 1/2 , and O 1 s at about 929.46, 955, and 525.86 eV, respectively [32].These results show that copper in metallic form has been successfully deposited on CNT.

SEM-EDX Analysis. Figures 4(a)-4(c)
show the SEM images of Cu loaded CNT.It is evident from the SEM of 10% and 20% Cu/CNT that the Cu is homogenously coated on the surface of the CNT.However, as the Cu loading is increased, Cu particles tend to form agglomerates and particle size increased.Agglomeration is quite apparent in Cu loading 50% in Figure 4(c).This is confirmed by particle size estimated from XRD.The EDX images of Cu/CNTs (Figure 5) reveal the presence of carbon and copper.The initial unknown peak represents the aluminum holder which is used in JEOL JSM-6460LV machine.The presence of metal elements such as Cu reveals the successful deposition of metals on CNT.
3.1.5.TEM Analysis.TEM image of the Cu supported CNT in Figure 6 shows its characteristic tube like structure with tube diameter in the range 4-10 nm as specified by suppliers.
It is obviously seen that Cu nanoparticles are homogeneously coated on the surface of CNTs and intimately attached to the CNTs.For 20% Cu loading, the particle sizes are found to be of 3-5 nm.These observations are in good agreement with the particle size estimated from X-ray diffraction data shown in Table 1.However, the TEM micrograph for 50 wt.%Cu on CNT shows agglomeration of smaller Cu particles to produce larger particles (50 nm).At high loading of copper metals, larger copper particles are expected to display low catalytic activity.
3.1.6.N 2 Adsorption Isotherms.Figure 7 shows the N 2 adsorption-desorption isotherms for pure and copper loaded CNTs.All these isotherms show a typical type 1 V adsorption model according to the IUPAC classifications.Table 2 shows the chemical and physical properties for the CNTs-supported Cu-catalysts including the Cu loading contents, BET surface area, and pore volume.The BET surface areas for the CNTssupported Cu catalysts range from 357 to 121.6 m 2 /g.The averaged crystallite size of Cu supported on CNTs (estimated on the basis of TEM image) ranges from 3.8 to 59 nm, which depends on the Cu loading content.As increasing the Cu content of the CNTs supported Cu catalysts, both BET surface areas and the pore volume decreased, accompanied with the increase of the size of the Cu particles.Generally, the decrease of the Cu dispersion or the increase of the particle size will lead to the decrease of the catalyst activity.From the above characterization results, it is concluded that morphologies and structure of the CNT-supported catalysts depend on the extent of Cu loading.This variation in properties is expected to have a significant impact on the electrocatalytic activity of the prepared catalysts.

Linear Sweep Voltammetry.
In this study initially 10% nominal Cu content was selected to prepare catalyst for ERC. Figure 8 shows the linear sweep voltammetry (LSV) curves in the 0 to −3.0 V range for 10% Cu/CNT deposited on carbon paper under different conditions.After the onset potential (i.e., the potential at which current shows its first nonzero value), there exists a linear relationship in the current and potential for the rest of the scan range with absence of any evident cathodic reduction peak.For both cases, there was no detectable physical change in the electrode.This shows that the carbon paper and catalysts were physically stable in working potential.In absence of CO 2 , entire current is due to production of H 2 by HER.But in CO 2 saturated electrolyte, two competing reactions occur: CO 2 reduction and H 2 evolution reaction.HER is an undesirable side reaction of ERC as it consumes electrical energy that was meant for ERC.Under the N 2 saturation, the current density is found to be 0.0045 A/cm 2 whereas, under CO 2 saturation, it shows current density of 0.008 A/cm 2 at −1.5 V. Difference in current density for the two cases arises from the contribution from the CO 2 reduction which takes place under CO 2 saturation.The observed higher current with CO 2 saturated solution is in line with some of the findings of previous studies [18,[33][34][35].However, there are studies reporting that the current density under CO 2 saturated condition was lower in the N 2 saturated solution [16,36,37].In our case the increased current density under CO 2 saturated condition shows that the CNT-supported Cu catalysts are catalytically active for the ERC.Cathodic current obtained in our studies is higher than the results reported in the available literature [11,14,18,37].This increase in current could be due to the excellent activity of the highly dispersed copper particles on high surface area CNT (400 m 2 /g).Similar results were obtained by Yamamoto et al. on porous activated carbon [23].
In order to study the effect of Cu content on the ERC and the optimum loading, we prepare catalysts with different metal loading on CNTs from 10 to 60% which were prepared and tested.Figure 9 shows the effect of copper loading on the current-potential curve.In theory, the catalytic activity increases with the metal content in the catalyst.More active sites provide better probability of reactants reacting.Therefore, cathodic current was found to increase with increase in copper loading till 20%, and with further increase in loading, as shown in Figure 9.However, with further increase in metal loading, current eventually started decreasing.For example, currents for 10% Cu/CNT and 20% Cu/CNT at −1.5 V were 0.004 A/cm 2 and 0.0175 A/cm 2 , respectively.At the same potential, 30% Cu/CNT showed 0.012 A/cm 2 current and further increase in loading has no impact on the current.Similar observation was made by Lee et al. in ERC copper loaded gas diffusion electrode [38].This can be explained by the TEM and XRD of the catalyst.Beyond 20% Cu loading the size of the particles increased coupled with active sites becoming increasingly inaccessible for the reactant due to the reduced surface area and micropore sizes because of pore blocking and agglomeration of the copper particles.Therefore, the optimum loading for the CNT catalyst prepared by homogeneous deposition and precipitation is 20%.

Chronoamperometry.
To study the long term performance of the catalyst, potentiostatic electrolysis was carried out at −1.7 V for 6000 s. Results are shown in Figure 10.The current of the ERC increases with increase in copper loading until it reaches 20% and then decreases with further increase in copper content.This is consistent with the results from the LSV.The result shows that copper supported on CNT is catalytically active for at least 6000 s.
Methanol concentration in the catholyte was measured by GC equipped with TCD.Prior to the GC analysis, methanol was extracted from the bicarbonate solution in ether.Faradaic efficiency (FE) for methanol production was measured by the following formula: where  is number of electrons involved in the reaction to produce the product and it is six for methanol generation,  is the Faraday's constant,  is moles of product formed,  is the total current in ampere, and  is the time in seconds.Figure 11 shows the Faradaic efficiency of methanol production as a function of potential.The highest FE for methanol formation from CO 2 was found to be 38.4% and it was achieved at −1.7 V.The products formed in the gas phase were not detected.

Conclusions
Copper-supported carbon nanotube catalysts were prepared by deposition-precipitation method.Linear sweep voltammetry and chronoamperometry results show that all the prepared catalysts showed electrocatalytic activity towards the electrochemical reduction of carbon dioxide to methanol.The characterization results further showed that all electrocatalysts are stable and catalytically active in the potential window of the electrochemical reduction.The degree of copper metal loading on the CNT had an impact on the performance of the synthesized catalysts.The catalyst with 20 wt.% loading of copper was found to be the best performing catalyst among all the catalysts prepared and tested.It showed the highest current density in chronoamperometry and LSV with low onset potential.Therefore, it is concluded that 20 wt.% copper metal loading is the optimum metal loading on CNT for the electrochemical reduction of carbon dioxide.This metal loading provides compromise between the amount of metal active sites for the reaction to take place and the increase in metal particle size due to agglomeration which greatly reduces the activity of the catalyst.Methanol was detected as the primary reduction product by gas chromatography.

Figure 3 :
Figure 3: XPS spectrum of 20%Cu/CNT for high binding energy for relevant C 1 s, Cu 2 p 3/2 , and O 1 s detailed XPS regions.

Figure 10 : 8 FaradaicFigure 11 :
Figure 10: The current responses for the supported catalyst under various loading under a constant potential of −1.7 V in CO 2 saturated 0.5 M NaHCO 3 solution.

Table 1 :
XRD data analysis results for the Cu/CNT catalysts.Crystallite size gives an idea about the size of the particles.

Table 2 :
Textural properties of the prepared catalysts.